Development of the UIUC Aero Testbed: A Large …...Development of the UIUC Aero Testbed: A Large-Scale Unmanned Electric Aerobatic Aircraft for Aerodynamics Research Or D. Dantsker,

Fluid Dynamics and Co-located Conferences AIAA 2013-2807 24-27 June, 2013, San Diego, CA 31st AIAA Applied Aerodynamics Conference

Copyright © 2013 by the authors. Published by the American Institute of Aeronautics and Astronautics, Inc., with permission.

Development of the UIUC Aero Testbed:A Large-Scale Unmanned Electric Aerobatic Aircraft for

Aerodynamics Research

Or D. Dantsker∗, Miles J. Johnson†, Michael S. Selig‡, and Timothy W. Bretl§

University of Illinois at Urbana–Champaign, Department of Aerospace Engineering, Urbana, IL 61801

This paper describes the development of a large 35%-scale unmanned aerobatic platform named the UIUCAero Testbed, which is primarily intended to perform aerodynamics research in the full flight regime. Thegiant-scale aircraft with a 105-in (2.7-m) wingspan and weight of 37 lb (17 kg) was constructed from a com-mercially available radio control model aircraft with extensive modifications and upgrades including a 12-kWelectric motor system that provides a thrust-to-weight ratio in excess of 2-to-1. It is equipped with an avion-ics suite that contains a high-frequency, high-resolution six degree-of-freedom (6-DOF) inertial measurementunit (IMU) that allows the system to collect aircraft state data. This information set can be used to generatehigh-fidelity aerodynamic data that can be used to validate high angle-of-attack flight-dynamic models. Col-laboration in this project also led the Aero Testbed to have the capability to fly fully- and semi-autonomouslyin order to conduct autonomous flight research. A literature review of aerobatic unmanned aircraft used forresearch is first presented. Then the background and motivations for developing this platform are discussed.This is followed by a description of the planning and development that was involved. Finally, initial test flightresults are presented, which include flight path trajectory plots of several aerobatic maneuvers.

Nomenclature

AV I = avionics integrationARF = almost ready to flyCOT S = commercial off the shelfCG = center of gravityDOF = degree of freedomEEG = electroencephalogramIMU = inertial measurement unitRC = radio control

I. Introduction

Numerous unmanned aircraft have been developed in recent years by universities for a variety of tasks. A rathersmall number of them have aerobatic designs allowing them to perform maneuvers that would be outside the scopeof the normal flight regime. Some of these aerobatic unmanned aircraft are used for aerodynamics research while themajority are used for guidance and control research. These aircraft largely vary in size and weight from having a 10-inwingspan and weighing a few ounces to having a 10-ft wingspan and weighing 30 to 50 lb.

The UIUC Applied Aerodynamics Group recently developed and began testing a large unmanned aerobatic aircraft,referred to as the UIUC Aero Testbed. This research platform is a highly-instrumented 35%-scale model of the full-scale Extra 260 aerobatic aircraft. The platform was developed from a commercial off the shelf (COTS) airframe andhas a 105-in wingspan and weighs 37 lb. It is powered by a 13-kW electric motor that provides the aircraft with a

∗Graduate Research Fellow, AIAA Student Member. [email protected]†Ph.D. Candidate, AIAA Student Member. [email protected]‡Associate Professor, AIAA Senior Member. [email protected]§Assistant Professor, AIAA Member. [email protected]

1 of 18

American Institute of Aeronautics and Astronautics

thrust-to-weight ratio of greater than 2-to-1. The Aero Testbed is primary used for acquiring aircraft state data forhigh-fidelity aerodynamics research, but it is also equipped such that it can be used for autonomous flight research.

This paper will first briefly examine other universities’ aerobatic unmanned aircraft. The paper will then discuss thebackground and motivations behind developing the UIUC Aero Testbed. After that is a description of the methodologyused in the planning and then the development of the Aero Testbed along with the specifications. Results from initialtesting will be presented and finally a summary is given.

Figure 1. A photograph of the UIUC Aero Testbed, which is an electric unmanned 35% scale Extra 260 aerobatic aircraft with a 105 inwingspan, a weight of 37 lb, and is instrumented with an avionics suite that includes a 6-DOF inertial measurement unit, GPS receiver, andpitot probe among other sensors.

A. Literature Review of Aerobatic Unmanned Aircraft Used for Research

Many universities have developed unmanned aircraft for research in a variety of fields. However, focus here is placedon fixed-wing aerobatic aircraft, as this is the area relevant to the aircraft developed. Aerobatic unmanned aircraft usedin research vary greatly in size, from a 10-in wingspan and weighing a few ounces to a 10-ft wingspan and weighing30 to 50 lb. Generally, smaller aircraft are flown indoors and larger aircraft are flown outdoors. These aircraft aresometimes developed for use in aerodynamics research but more often end up being used as test vehicles for hardwareand/or control-scheme algorithms.

When performing research with miniature unmanned aerobatic aircraft, state data is collected externally usingmotion capture systems set in indoor flying spaces with no wind. For aircraft that are flown autonomously, the positiondata is used to close the loop in the control schemes. This was the case with researchers at the Georgia Institute ofTechnology, where a miniature indoor aerobatic aircraft was used to demonstrate a guidance controller for transitionsinto hovering state.1 Similar work was done at the Massachusetts Institute of Technology2 and Universit Laval.3 Usinga similar motion capture system, researchers at the University of Illinois parameterized a micro, aerobatic aircraft todetermine its aerodynamic characteristics, including lift, drag, and moment curves over a wide flight regime.4

Moving onto outdoor aircraft, researchers at the University of Florida used a small outdoor aerobatic unmannedaircraft to investigate aircraft high angle-of-attack flight dynamics, which included stability issues that related toside slip5 and wing rock6–8 and later further parameterized the same aircraft in order to design trajectory-followingalgorithms;9, 10 this aircraft had all sensors and recording equipment on board. However, performing experimentsoutdoors does come with the problem of external disturbances, the most significant of which is wind. In general, thelarger the aircraft, the less the aircraft will be affected by a given amount of external disturbance. Larger aircraft alsohave the advantage that they are able to carry more weight, without a significant decrease in their flight performance.This is the reason why researchers in ETH Zurich choose the size they did; they used a 10.2-ft wingspan, 61-lb,unmanned aerobatic aircraft to perform a variety of aerobatic maneuvers autonomously.11 Similar research was doneat Stellenbosch University.12, 13 At the Georgia Institute of Technology, researchers used a large, 8.75-ft wingspan,35-lb, aerobatic unmanned aircraft to perform a variety of tasks including autonomous transitions to hover.14, 15 They

2 of 18


also used this aircraft as the flight leader in a vision-based formation flight with a large unmanned helicopter16 andwith another large, 8.5-ft wingspan, 28-lb, aerobatic aircraft.17

There are several examples of large aerobatic unmanned aircraft being used for risk mitigation research. Re-searchers at NASA ARC and LaRC used a large scale electric aerobatic aircraft with an 8.25-ft wingspan to performresearch in structural fault diagnostics and mitigation, and in electric motor battery health management.18 Work at theUniversity of Illinois using a 6.5-ft wingspan aerobatic unmanned aircraft involved risk mitigation in the form of failuremode analysis of subsystems and components,19 particularly fuel management,20 system fault detection,21 and objectand terrain detection using optical flow.22, 23 The aircraft used by the University of Illinois was also used by Boeing totest distributed communication networks.24 The University of Kansas has performed extensive testing with their large,10-ft wingspan, aerobatic unmanned aircraft: researchers have evaluated a COTS autopilot,25, 26 tested a new flightcontrol system,27, 28 performed aircraft and avionics system identification,29–31 used it as the base aircraft from whicha test pilot transitioned to a new unfamiliar airframe,32 evaluated flight loads using strain gauge measurements33 andcompared moment of inertia estimation methods with experimental measurements.34

II. Background and Motivations

Figure 2. Simulated spiral maneuver executed by the Aero Testbedin FS One V2.

The Aero Testbed concept began with members ofthe UIUC Applied Aerodynamics Group striving to pos-sess an aerobatic aircraft platform that could record high-fidelity aerodynamic data. This platform would allow forresearch to be done in the full-envelope flight regime,that is, over the full +/-180 deg range in angle of at-tack and sideslip. It would be able to record high-frequency, high-resolution aircraft state data in regularaerobatic and high angle of attack flight, outside therange of conventional aircraft and in maneuvers onlyachievable by recently developed unmanned aircraft andhigh-performance radio control (RC) model aircraft. Thestate data that this aircraft would record includes threedimensional linear and angular accelerations, velocities,and positions along with airspeed, control surface de-flections, and engine performance. Such data couldthen be used in the validation of aerodynamics meth-ods applied to aircraft stall/spin and upset scenarios, e.g.,methods like those used in the real-time flight simulatorFS One.35, 36 In order to develop such an aircraft, manydecisions would need to be made concerning the desired aircraft performance in terms of airframe and instrumentationspecifications.

The UIUC Applied Aerodynamics Group eventually collaborated on this project with the UIUC Robotics andNeuro-Mechanical Systems Laboratory. The Applied Aerodynamics Group was responsible for fabricating the air-frame while the Robotics and Neuro-Mechanical Systems Laboratory would develop and integrate an avionics suiteinto the aircraft using the knowledge gained from developing the Fixed Wing Multi-Role Unmanned Aircraft VehicleResearch Testbed along with previously developed unmanned aircraft.37 In terms of capability, the aircraft evolvedinto being a dual-purpose unmanned aerial vehicle. When the Aero Testbed would be used by the Applied Aerodynam-ics Group, the aircraft would be piloted from the ground using a remote control transmitter, manually performing alltypes of maneuvers. In the other flight configuration, while it would be flown by the Robotics and Neuro-MechanicalSystems Laboratory, the Aero Testbed would also be used for doing fully- and semi-autonomous flight. In that case,research conducted would include human interface flight, specifically using a neural interface, and autonomous flightsuch as flying aerobatic patterns, taking-off, and landing using a perching technique. Research using neural interfacesfollows several other unmanned aircraft that were successfully flown/controlled by a pilot using an electroencephalo-gram (EEG) as input and live onboard video as visual feedback.38, 39 All autonomous research would be done usingthe onboard avionics package in autopilot mode, which would control the aircraft. It is important to note that therewould be a pilot capable of remotely taking over control if necessary, via a toggle switch on the transmitter.

3 of 18


III. Planning

In order to develop an unmanned aerobatic aircraft that would serve well in all expected roles, careful planningwas required. Planning through all stages of construction and instrumentation and examining how the aircraft wouldbe used prevented or at least diminished the number of problems that arose. This section is separated into severalsub-sections, which cover airframe and component selection through proper flight testing.

A. Airframe Size and Construction

Figure 3. A photograph of Eli Field (taken from the Monticello ModelMasters RC Club40).

The size of the aircraft that would be developed wouldneed to be determined. In general, a given payloadweight will affect a larger aircraft less than it would asmaller one because the percentage weight increase, andtherefore the loading increase, will be less due to thehigher starting weight. Also, the larger the aircraft, theless it would be affected by environmental conditionssuch as wind. Increasing the size also makes it easierto work inside the aircraft. However, having a large air-craft does produce a variety of problems and limitations.Larger aircraft are more costly to build and operate thansmaller aircraft and require a larger vehicle to transportthem from the laboratory to the flying site. The nearestflying site that could facilitate the flight tests is locatedapproximately 25 mi (40 km) away from the Universityof Illinois campus; this flying site is Eli Field, which isowned and maintained by the Monticello Model MastersRC Club.40 It should be noted that in order to save valuable time in developing this platform, a COTS airframe wouldbe used, which creates a limitation in terms of available models and their sizes. After considering these factors, it wasdetermined that the most appropriate aircraft size range was 33–40% of full-scale aerobatic aircraft, which yields awingspan range of 7.5–9 ft (2.3–2.7 m).

Aircraft airframes in the desired size range are predominantly constructed in one of two ways. They are eitherbuilt-up of plywood structures, sheeted with balsa, and then covered with a light plastic shrink wrap, or they aremolded out of composites, which include carbon fiber, fiberglass, honeycomb, and sandwich of composite cloth andlight-weight filler. In terms of pros and cons for each building technique, composite construction generally is lighter,yields more rigid structures, and can produce more accurate designs. Composites, however, are more expensive anddifficult to repair. Wood built-up designs are relatively low cost, can easily be built, repaired, and modified but are notas rigid. Wooden designs also have the unique problem of bowing when exposed to a significant amount of sunlight,especially if the aircraft is covered by a dark-color film. Bowing is caused by the plastic film covering the aircraft:after being exposed to sunlight for a sufficient amount of time, the film heats up and contracts, which is a problemsince only one half of the aircraft receives sunlight. In most cases, the film on the top side of the aircraft contractsand bows the wings and horizontal tail up. It should also be noted that there may be combinations of the two methodsavailable, depending on the airframe, such as a composite fuselage with wooden built-up wings and tail pieces. Giventhe choice of construction methods, the wooden built-up structure was chosen for its versatility, which in retrospectprovided the ability to modify and build off of internal structural elements that became critical in instrumenting theaircraft.

4 of 18


B. Airframe Propulsion

The next decision in planning the aircraft was what type of propulsion system should be used. There were severalchoices for an airframe of this size. The default choice was using the gasoline engine recommended for the airframechosen or a slightly larger one, or on the other hand a large electric motor could be used. There are numerousadvantages for each type of propulsion system, in terms of both practicality and data collection. Operational safetymust also be considered.

(a) (b)

Figure 4. Examples of aircraft propulsion system options, at approximately the same scale: a) on the left is a Desert Aircraft DA-120gasoline engine and b) on the right is a Hacker A150-8 electric motor (taken from Ref. 41 and Ref. 42, respectively).

First of all, using an electric motor creates much less vibrations through the aircraft. Having less vibration beingtransmitted throughout the aircraft decreases the noise recorded by the instruments, thereby yielding better data. Inaddition, having less vibrations traveling through the airframe will lower the rate of fatigue and thus increase the lifeof the airframe and avionics. Also, using an electric system results in no center of gravity (CG) shift during flight orweight change from fuel being burnt. Having the aircraft remain the same from takeoff to landing is important factorin examining aerodynamic data because the airframe will always have constant inertial properties throughout the flightand day-to-day. With an electric system, the input power can be measured precisely, and it is much easier to measurepropeller speed (RPM) by measuring the current running through poles of the motor instead having to perform opticalmeasurements, which are often difficult in changing light conditions. Moreover, in a practical sense, electric motorshave constant performance and need not be tuned, they have much less lag in terms of throttle response than a gasengine, are much easier to install, and have few moving parts, which means they also need less maintenance, and ofcourse they run clean (i.e., the airframe does not become covered with contaminates such as oil and fuel).

Gasoline engines are preferable to electric motors in several categories. First, all airframes in the size range desiredare designed for gas engines, which means to use an electric motor requires modifications such as constructing a motormount and battery trays. Also, an aircraft equipped with a gas engine generally has greater endurance than with anelectric motor due to the fact that the fuel required for a given flight time weighs far less than the equivalent amount ofbatteries. Furthermore, the weight of a gas engine is the same as an electric motor with onboard equipment; however,in terms of gasoline versus batteries, batteries generally yield a 10% increase in the weight of a flight ready aircraftover one that would be fully fueled. Otherwise, electric motors pose a variety of challenges that gas engines simplydo not, the most prominent of which is than electric motors use high voltages and currents. For an aircraft in the sizerange purposed, the motor system at full throttle would run at 50–60 VDC and 200–250 Amps, or about 10-15 kW. Inaddition, the lithium polymer (LiPo) batteries used are highly volatile and must be handled with care.

In terms of cost between the two options, gas offers pay-per-use while electric propulsion requires that everythingis paid upfront. More specifically, the cost of each option with all the required accessories and the peripheral equipmentis equal, less the cost of fuel for gas and batteries for electric. However, with the electric system, batteries must bebought ahead of time, and the number of motor battery packs will equal the number of flights that can be flown.Yet, if the motor battery packs are charged at the flying site, the number of flights per day can be increased. Withregards to operational cost, given current fuel prices, the cost of one motor flight pack is approximately equivalent toapproximately 500 flights worth of gas. This figure does ironically correlate to the optimal number of cycles, or flights,that can be achieved with a battery; however, if battery life is inadvertently shortened, then gas is less expensive.

The last major factor that must be considered is safety. Gas engines require manual starting, specifically by turningthe propeller over by hand, which is a major safety concern. However, when a gas engine is off it cannot start bymistake; whereas, whenever the batteries are connected into the motor system everything is armed. To solve the

5 of 18


issue of the motor system always being armed when the batteries are plugged in, an inline electronic switch with a“remove-before-flight” indicator can be added to make the state of the motor obvious.

Given all the aspects considered, using an electric motor system was chosen mainly due to the superior aerody-namics data that the platform could yield, which was one of the fundamental reasons for starting the project. Anothermajor reason is that overall it is a more practical system to use and requires less maintenance.

C. Airframe Choice

Provided the choices made for size, airframe construction, and propulsion system, after examining several COTSairframes, the Hangar 9 35%-scale Extra 26043 was chosen. The 35%-scale Extra 260 is a relatively light and strongdesign that, once instrumented, has a similar weight to the un-instrumented weight of other airframes in the samesize category. In order to convert the airframe an RC model aircraft into an unmanned aircraft, some of the stockhardware needed to be upgraded. Given that the airframe would be instrumented and carry valuable equipment,it is advantageous to improve the overall safety and performance characteristics. These hardware changes includeupgrading the servos to stronger and faster models, upgrading all fasteners and linkages, and of course reinforcing thestructure. Also a power distribution system would be added, not only to manage power flow sent to servos but to makethe flight control system power redundant.

Since the propulsion system was chosen to be electric, certain modifications needed to be made. These modifica-tions include properly mounting the motor and widening the support structure to compensate for the greater torque anelectric motor could create compared with a gas engine. Also, battery trays needed to be fabricated and installed suchthat they provide a sufficient amount of room to change the CG of the aircraft, either to aid the pilot in performingcertain maneuvers or to optimize the aircraft for autonomous control.

Figure 5. Stock photograph of the Hangar 9 35%-scale Extra 260 (taken from Ref. 43).

D. Instrumentation

There are essentially two parts making up the platform—the airframe being one and the avionics suite being thesecond. The avionics suite must be able to perform two tasks: record flight state data while the aircraft is being flownmanually, and fly the aircraft either fully- or semi-autonomously. In the case of autonomous flight, the safety pilot atthe field must be able to regain control at any time. Otherwise, the goals in selecting an avionics system would to beto maximize safety, maximize logging rate and resolution, minimize complexity, and minimize cost. Several optionswere discussed during the planning stage. First of all, the method used to operate the avionics in each flight task wasexamined. One option was to use a stand-alone autopilot and a stand-alone data logger and to run one while the other

6 of 18


one is off. The next option was to use an autopilot capable of both flight and logging and to just disconnect the autopilotduring manual flight. Either case could be done with COTS hardware or by developing custom electronics. It shouldalso be noted that for any of these operational options a method for reliably switching from autonomous-to-manualcontrol needed to be devised. In the end, the avionics suite chosen to be used on the aircraft was an upgraded versionof that developed and used on the Fixed-Wing Multi-Role Unmanned Aircraft Vehicle Research Testbed,37 with theaddition of a separate COTS system to monitor the electric propulsion system.

E. Flight Testing

An important factor in the development of an unmanned testbed platform is to follow a procedure that allows for allof the systems to be tested while minimizing risk to the other systems. In this case, that meant being able to testthe airframe and avionics while minimizing the endangerment of the other. The plan for testing the Aero Testbedwas extensively discussed and emerged as the following. The first step to test the platform would occur once theairframe was constructed and flight-readied as an RC aircraft. Then it would be test flown several times to observeif any problems arose. This means that the RC test pilot would run the aircraft through its paces. If any problemswere detected, the appropriate modifications would be made. Then the next step would be to put the avionics systemtogether in the aircraft and perform diagnostic tests. Assuming everything checked out, the aircraft would be testflown with the system turned off. Then it could be flown with the avionics running in recording mode. Finally forautonomous flight, the flight testing / risk mitigation method used with the Fixed-Wing Multi-Role Unmanned AircraftVehicle Research Testbed37 would be adopted.

IV. Development

The development process of the Aero Testbed is separated into two stages: aircraft construction and then instru-mentation integration. The completed aircraft specification are presented.

A. Airframe Construction

The construction of the Aero Testbed airframe started with a Hangar 9 35%-scale Extra 260 almost-ready-to-fly (ARF)RC airplane kit. The kit came with all major components of the airframe prefabricated along with a variety of hardwarefrom landing gear components to linkages. Generally with ARF airplane kits, a good deal of assembly is required tobring the airplane to flying condition; however, ARF kits are much less time consuming compared with traditional kitswhere everything, including the structure, must be constructed and then assembled.

Figure 6. All major airframe components that arrived with Hangar 9 35%-scale Extra 260 ARF kit.

As described in the planning section, most of the hardware included with the kit was replaced to better adaptthe airframe for use as an unmanned testbed platform. These changes were made to improve flight and longevity

7 of 18


characteristics, as well as an overall safety consideration. Most of the stock fasteners included with the airframewere replaced with identically sized fasteners made from stronger materials in order to minimize the risk of failure.Another upgrade included the replacement of the aluminum landing gear with a pair of Graph Tech RC44 carbon-fiberlanding gear that could better handle landings of an airplane that is above its original design weight. The main axleswere also replaced in this fashion. The wing and horizontal stabilizer joiner tubes were replaced with a pair of hightensile strength carbon-fiber wrapped tubes. Other upgrades included the replacement of the linkage connectors withheavy-duty Dubro45 connectors and the steel linkages with stronger titanium linkages. To facilitate the installation ofthe flight control system components and the avionics, two large balsa/carbon-fiber sandwich trays were fabricatedinstalled in the fuselage. These trays also served to strengthen the area that would experience increased load.

Figure 7. Smart-Fly power distribution unit, JR receiver, and 2ircmanual overrides installed on the front balsa/carbon-fiber sandwichtray.

With regard to the RC flight control system, severalupgrades and additions were made from what would nor-mally be used on a normal 35%-scale Extra 260. The ser-vos installed for the flight controls were at the top of themanufacturer’s range of recommended servos. It shouldbe noted that there are two servos for each aileron andfor the rudder, and a separate servo for each of the el-evators, to increase redundancy in the system. The RCreceiver used (JR R1221X) is the longest range modelavailable, uses 2.4-GHz frequency hopping modulation,and has four redundant antennas. The controller usedwith this receiver is the JR 12X.46 A manual overrideswitch was installed to allow shifting control betweenthe RC receiver (that the human pilot would use) and theautopilot. It should be noted that this is a COTS stand-alone unit made by 2icrc that is controlled by a servochannel from the RC receiver, which the test pilot al-ways has control over.47 A Smart-Fly power distributionunit was installed in the aircraft to both filter and regu-late the signals and power available to each servo. It alsoserved to duplicate the servo channels so that they can berecorded; the unit also acts as a redundant power supplyby means of two batteries.48 The two batteries used topower the flight control system are Thunder Power 2S (2cell / 7.4 V nominal) 5-Ah LiPo batteries.49

The Aero Testbed is equipped with an electric motor system consisting of a Hacker A150-8 brushless motorrunning a Mejzlik 27x12TH electric thin-blade propeller50 and a Hacker MasterSPIN 220 electronic speed controller(ESC).42 The motor is connected to the speed controller that regulates the rotation rate of the motor by pulsing themotor coils. An EMCOTEC Safety Power Switch was installed in-line between the electronic speed controller andthe flight battery and allows the flight battery to be connected without the motor engaging. The switch is triggeredusing a magnetic pull “Remove Before Flight” flag which was installed on the top of the cowling for easy visibility.51

The flight pack motor batteries were assembled from four Thunder Power 7S (7 cell / 25.9 V nominal) 5-Ah LiPobattery packs in a 2S 2P configuration, meaning that two packs in series are paralleled with a second set of two packsin series. The flight pack is configured as a 14S (14 cell / 51.8 V nominal) 10-Ah LiPo battery pack. All battery-levelinterconnects are made using 10-AWG wire and 4-mm gold-plated bullet connectors and run up to 125 A at 29.4 V.All motor power connections, from the switch to the motor are made using 8-AWG wire, or a set of three 12-AWGwires in parallel, and 6-mm gold-plated locking bullet connectors, and run up to 250 A at 58.8 V. The entire systemoperates using direct current. In order to achieve the proper CG, battery trays were installed in the aircraft where thefuel tank originally sat. The trays were oversized in order to make it possible for the CG to be adjusted.

8 of 18


Figure 8. A propulsion system diagram for the Aero Testbed unmanned aircraft.

An Eagle Tree Systems Flight Data Recorder (FDR) was fitted on the Aero Testbed to monitor the propulsionsystem and the overall state of the aircraft.52 The Eagle Tree Systems FDR monitors the voltage of the motor flightpack, voltage output of the power distribution unit, current drawn by the motor, motor RPM, temperature of the motor,electronic speed controller, and all motor batteries. Additionally, the flight control inputs sent out from the powerdistribution unit and the GPS position of the aircraft are recorded. The FDR records all these values at a rate of 10 Hzbut can be configured to run at up to 40 Hz. It transmits the information it logs to a ground station via a 2.4-GHztransmitter at a slightly lower rate. The Eagle Tree Systems FDR was primarily installed to allow the propulsionsystem to be monitored.

Figure 9. The Hacker A150-8 brushless motor (bot-tom right), MasterSPIN 220 electronics speed controller(top middle), and SPS safety switch (center left) seen be-fore the cowling, propeller, and spinner were installed.

Figure 10. A photograph on the motor battery trayswhere the Eagle Tree Systems Flight Data Recorder canbe seen at the top of the shot.

9 of 18


B. Avionics Integration

The Aero Testbed was equipped with an avionics suite based on the system developed and used on the Fixed-WingMulti-Role Unmanned Aircraft Vehicle Research Testbed.37 The avionics suite includes a Gumstix flight computer,53

an ARM7 microprocessor based Paparazzi autopilot,54 and a Xsens MTi-G six degree-of-freedom (6-DOF) inertialmeasurement unit (IMU),55 along with several other components. This system was first built and tested outside of theaircraft and then was reassembled and mounted in the aircraft.

Paparazzi

TWOG

Autopilot

Gumstix

Overo

Flight Computer

IMU

GPS

Pitot 900 MHz

TX/RX

2icrc

Multiplexer

Motor Servos

Camera

5.8 GHz

TX

Eagle Tree FDR

2.4 GHz TX

Temp

RPM

Voltage

Signal (arrow):

█ Flight controls

█ Multiplex override

█ Sensor data

█ System state data

█ Transmission data

Power (shading):

█ Flight controls

█ Motor

█ Instrumentation

Color Key

Smart-Fly

PDU

Futaba R/C

2.4 GHz RX

Figure 11. A flowchart diagram of the avionics suite showing a top-down view of flight control signals, multiplex the override commandsignal, sensor inputs, instrumentation communication, and transmission routes.

The IMU is a MEMS-based 6-DOF inertial measurement unit that has an internal barometric pressure sensor andan internal GPS receiver along with two 16-bit analog-to-digital converters. The IMU has an onboard Attitude andHeading Reference System (AHRS) and Navigation processor that runs a real-time sensor-fusion algorithm using aproprietary extended Kalman filter to output minimal drift, GPS-enhanced, 3D orientation and position data at a rateof up to 120 Hz. The IMU uses this filter to couple inertial measurements from the accelerometer, gyroscope, andmagnetometer that update at 512 Hz with a 4-Hz GPS and a static pressure sensor that updates at 9 Hz. The GPSreceiver is connected to an antenna that was mounted on top of the fuselage behind the canopy hatch area. GPSaltitude readings are aided by the barometric pressure sensor. The IMU is also able to output all the unprocessedsensor readings, which is useful because body-frame accelerations and rate-of-turn can be logged. The airspeed ofthe aircraft is measured by a pitot probe connected to an All Sensors 20cmH2O-D1-4V-MINI differential pressuresensor,56 which is integrated into the system through one of the two IMU analog-to-digital converters.

The flight computer and autopilot continuously communicate over a serial line. The flight computer was setup suchthat it records aircraft state data independent of whether or not the aircraft is being flown manually or autonomously.When the aircraft is being flown autonomously, it performs all the high-level flight-plan calculations. The autopilot onthe other hand is used primary as input/output device in that it passes data from the IMU to the flight computer and usesthe flight plan commands from the flight computer to run low-level control loops to output the proper control-surfaceservo commands via its servo output ports. The flight computer and autopilot operate at a rate of 100 Hz; however,the autopilot outputs servo commands at a rate of 60 Hz. The avionics suite is also equipped with an XTend 900-MHztransceiver,57 which is connected to the flight computer via a serial line, allowing the avionics system to communicate

10 of 18


with a ground station where the ground-station operators can both monitor acquired data and control the aircraft ifneeded.

Physically, the majority of the avionics system, which includes the flight computer, autopilot, autopilot-to-IMUsignal conversion board, avionics power regulators, and avionics batteries, were mounted on the rear balsa/carbonfiber tray. The avionics are supplied with power from two regulators—the first made by Smart-Fly redundantly inputspower from two batteries and lowers the voltage to 6.5 V and the other made by Castle Creations,58 which is connectedto the first regulator and lowers the voltage further to 5 V. Power originates from two Thunder Power 2S (2 cell / 7.4V nominal) 900-mAh LiPo batteries.

The remainder of the avionics components lay throughout the rest of the aircraft. The IMU was mounted as closeas possible to the CG (right above the wing tube of the aircraft). This placement was done in order to minimizethe accelerations created by aircraft rotation, which is important in an aircraft that is able to produce large angularaccelerations. The pitot probe was mounted on the left wing tip, while the differential pressure sensor was mountedhalfway through the wing. The avionics transceiver was mounted halfway into the tail to minimize possibilities ofinterference with other equipment.

After several test flights with the aircraft instrumented, a passive high-definition video camera was added to thecanopy of the aircraft in order to record a pilot’s eye view. This video greatly aids in recognizing maneuvers and theirtiming during the post processing of data. In order to facilitate human-machine interface-based autonomous flight, theaircraft has space available for a gimbaled video camera that can be downlinked using a 5.8-GHz transmitter, allowinginterface subjects to have point-of-view visual feedback.

Figure 12. The avionics suite components, starting on the top left and going clockwise: a) the IMU is mounted as close as possible to theCG, b) the pitot probe is mounted on the left wingtip, c) the GPS antenna is mounted atop the fuselage behind the canopy, and d) the rearbalsa/carbon fiber sandwich tray with the flight computer, autopilot, IMU signal conversion board, and two regulators.

11 of 18


C. Completed Specifications

The completed specifications of the UIUC Aero Testbed are given in this section. First the aircraft physical specifica-tions are given in Table 1. Then the airframe specifications are given in Table 2. Finally the avionics specificationsare given in Table 3.

Table 1. Aircraft Physical Specifications

Geometric PropertiesOverall Length 98 in (2489 mm)Wing

Span 105 in (2667 mm)Area 2003 in2 (129 dm2)

Airfoil Symmetric - 10.5% ThickAspect Ratio 5.50Taper Ratio 0.45Dihedral 0.00 degIncidence 0.00 degSweep < 1 degAileron Area 20.0% of Wing Area

Horizontal StabilizerArea 449 in2 (29.0 dm2)

Airfoil Symmetric - 11% ThickAspect Ratio 4.90Taper Ratio 0.55Dihedral 0.00 degIncidence 0.00 degElevator Area 45.4% of Horizontal Stabilizer Area

Vertical StabilizerArea 214 in2 (13.8 dm2)

Airfoil Symmetric - 11% ThickRudder Area 72.0% of Vertical Stabilizer Area

Inertial PropertiesWeight

Empty (w/o Battery) 31.3 lb (14.2 kg)14S LiPo Battery 6.2 lb (2.8 kg)Gross Weight 37.5 lb (17.0 kg)

Wing Loading 40.3 oz/ft2 (123 gr/dm2)

CG Location 35.8% of Wing Mean Aerodynamic ChordMoments of Inertia

Roll 1.13 slug-ft2 (1.53 kg-m2)

Pitch 3.58 slug-ft2 (4.86 kg-m2)

Yaw 4.47 slug-ft2 (6.06 kg-m2)

12 of 18


Table 2. Airframe Specifications

Construction Built-up balsa and plywood structure, foam turtle deck, carbon fiber wingand stab tube, carbon fiber landing gear, fiberglass cowl, fiberglass wheelpants, and styrene canopy.

Flight ControlsControls Aileron, elevator, rudder, and throttleTransmitter JR 12X DSM2Receiver JR R1221 DSM2Servos (8) JR DS8711 Ultra TorquePower Distribution SmartFly PowerSystem Competition 12 TurboReceiver Battery (2) Thunder Power ProLite 20c 2S 5000 mAh (redundant)

PropulsionMotor Hacker A150-8 OutrunnerESC Hacker MasterSPIN 220 Brushless Speed ControllerPropeller Mejzlik 27x12THSpinner TruTurn 4.5-in Ultimate with Lite Backplate & Blue Anodize Alum.Motor Flight Pack (4) Thunder Power ProPower 30c 7S 5000 mAh in 2S2P config.Motor Power Switch Emcotec SPS 120/240Static Thrust & RPM 76 lb at 6,936 RPM at standard day at 770 ft altitudeStatic Thrust / Weight 2.05:1 ratio

Flight Time 8-12 min

Table 3. Avionics Specifications

Onboard SystemsFlight Computer Gumstix Overo Water with Summit expansion boardAutopilot Paparazzi TWOG v1.0

SensorsIMU XSens Mti-g 6-DOF IMU with Wi-Sys WS3910 GPS AntennaAirspeed Pitot-static probe using an All Sensors 20cmH2O-D1-4V-MINI differen-

tial pressure sensorTransceiver Digi 9X Tend 900-MHz cardPower

Regulators Smart Fly SportReg & Castle Creations CCBECBatteries (2) Thunder Power ProLite RX 2S 900 mAh (redundant)

Data Storage MicroSD card up to 8 GBData Log Rate Up to 100 Hz

13 of 18


V. Initial Flight Test Results

During the spring, summer and fall of 2012, the Aero Testbed was flown in its fully-instrumented configurationseveral dozen times. Initially, simple circuits were flown, and as the flight days progressed, maneuvers were addedto the flight plans as a higher level of familiarity with the aircraft and the safety procedures were gained. The AeroTestbed was flown through a variety of aerobatic maneuvers, specifically including spins, which became of particularinterest.59, 60 These maneuvers were performed to test the avionics system because of their relative simplicity andsafety compared with more complicated maneuvers. The avionics were initially set to log at 5 Hz, and then this wasincreased to 25 Hz. A handful of rudimentary aerobatic maneuvers that were performed are shown in in Fig. 13.As can be seen in several of the trajectories, the flight paths become jagged when the aircraft attitude is rapidlychanging. The jaggedness is the result of filtering problems occurring in the IMU processor due to temporary GPS andmagnetometer sensors faults following the rapid course changes. These issues will be solved by performing additionalpost-processing of raw sensor data.

of 18

!"#

!##

"#

#

!"#

!##

"#

#

"#

!##

!"#

$##

$"#

%&'()*&'()

+&'()

!""

#"

$"

%"

&"

"

&"

&&"

&""

!#"

!$"

!%"

!&"

"

&"

%"

$"

#"

!""

!&"

!%"

!$"

!#"

&""

'()*+,()*+

-()*+

!

"!!

" !

#!!

# !

$%!!

$& !

$&!!

$# !

!

!

"!!

" !

'()*+,()*+

-()*+

!" #"

$" %"

&" '"

$" %"

&" '"

("

"

%

)"

)%

("

(%

*+,-./+,-.

0+,-.

Figure 13. Trajectory and ground trace plots of aerobatic maneuvers, starting on the top left and going clockwise: a) loop, b) Immelmann,c) roll, and d) half Cuban-eight.

14 of 18


of 18

!"# !$#

!%#

!#%#

$#"#

&##&!#

&%#

#

!#

%#

$#

"#

&##

&!#

&%#

&$#

&"#

!##

'()*+

,()*+

-()*+

−150

−100

−50

0

−80−60

−40−200

0

50

100

150

200

250

300

350

400

450

x (m)y (m)

z (

m)

Figure 13. [continued] e) spilt-S and f) spin

VI. Summary

This paper described the development and initial testing of the UIUC Aero Testbed, which is a large unmannedaerobatic aircraft primary intended to perform aerodynamics research in the full flight regime. The aircraft, whichhas an 105-in wingspan and weighs 37 lb, is powered by a 12-kW electric motor that provides the aircraft with athrust-to-weight ratio in excess of 2-to-1. The aircraft is equipped with an avionics suite that can be used to generatehigh-fidelity aerodynamic data using a high-frequency, high-resolution 6-DOF IMU. This aerodynamic data will usedto validate high angle-of-attack flight-dynamics models along with allowing stall/spin upset conditions to be studied.To date, the Aero Testbed has flown several dozen times, each time with an increasing amount of rudimentary aerobaticmaneuvers. Once the logging and processing of these maneuvers can be perfected, more complicated maneuvers willbe performed and analyzed, particularly in the high angle-of-attack regime.

15 of 18


Figure 14. The Aero Testbed flying inverted.

Figure 15. The Aero Testbed landing.

Acknowledgments

We gratefully acknowledge the members of the UIUC Applied Aerodynamics Group: Gavin Ananda, Giovanni Fiore,Brent Pomeroy, Adam Ragheb, Agrim Sareen, and Pritam Sukumar for their support during flight testing, and Robert De-ters and Daniel Uhlig for stimulating discussions. We would also like to acknowledge Abdullah Akce, Mahsa Kasraie,Ben Parks, and Debra Reitz for their help during construction, and Ayo Adesida, Gustavo Fujiwara, and Craig Paugafor their support during flight testing. The authors would like to thank the UIUC Robotics and Neuro-MechanicalSystems Laboratory for sharing the testing resources used in this research, and the UIUC AE Machine Shop fortheir help, especially Greg Milner and Stephen Mathine. We would also like to thank Horizon Hobby, specificallyMichael McConville for providing design data and CAD drawings for his Extra 260 aircraft design and David Ribbefor initial aircraft checkout and for first test flights, and the Monticello Model Masters RC Club for allowing the useof Eli Field.

16 of 18


References1Blauwe, H. D., Bayraktar, S., Feron, E., and Lokumcu, F., “Flight Modeling and Experimental Autonomous Hover Control of a Fixed Wing

Mini-UAV at High Angle of Attack,” AIAA Paper 2007-6818, AIAA Guidance, Navigation, and Control Conference, Hilton Head, South Carolina,Aug. 2007.

2Frank, A., McGrewy, J. S., Valentiz, M., Levinex, D., and How, J. P., “Hover, Transition, and Level Flight Control Design for a Single-Propeller Indoor Airplane,” AIAA Paper 2007-6318, AIAA Guidance, Navigation, and Control Conference, Hilton Head, South Carolina, Aug.2007.

3Bilodeau, P. R., Poulin, E., Gagnon, E., Wong, F., and Desbiens, A., “Control of a Hovering Mini Fixed Wing Aerial Vehicle,” AIAA Paper2009-5794, AIAA Guidance, Navigation and Control Conference, Chicago, Illinois, Aug. 2009.

4Uhlig, D., Sareen, A., Sukumar, P., Rao, A. H., and Selig, M. S., “Determining Aerodynamic Characteristics of a Micro Air Vehicle UsingMotion Tracking,” AIAA Paper 2010-8416, AIAA Guidance, Navigation, and Control Conference, Toronto, Ontario, Canada, Aug. 2010.

5Perry, J., Mohamed, A., Johnson, B., and Lind, R., “Estimating Angle of Attack and Sideslip Under High Dynamics on Small UAVs,”Proceedings of the ION-GNSS Conference, Savannah, Georgia, 2008.

6Johnson, B. and Lind, R., “High Angle-of-Attack Flight Dynamics of Small UAVs,” AIAA Paper 2009-61, AIAA Aerospace SciencesMeeting, Orlando, Florida, Jan. 2009.

7Johnson, B. and Lind, R., “Characterizing Wing Rock with Variations in Size and Configuration of Vertical Tail,” AIAA Paper 2009-6151,AIAA Atmospheric Flight Mechanics Conference, Chicago, Illinois, Aug. 2009.

8Johnson, B. and Lind, R., “Characterizing Wing Rock with Variations in Size and Configuration of Vertical Tail,” Journal of Aircraft, Vol. 47,No. 2, 2010, pp. 567–576.

9Johnson, B. and Lind, R., “3-Dimensional Tree-Based Trajectory Planning with Highly Maneuverable Vehicles,” AIAA Paper 2011-1286,AIAA Aerospace Sciences Meeting, Orlando, Florida, Jan. 2011.

10Johnson, B. and Lind, R., “Trajectory Planning for Sensing Effectiveness with High Angle-of-Attack Flight Capability,” AIAA Paper 2012-0276, AIAA Aerospace Sciences Meeting, Nashville, Tennessee, Jan. 2012.

11Mockli, M., Guidance and Control for Aerobatic Maneuvers of an Unmanned Airplane, Ph.D. thesis, ETH Zurich, Department of Mechanicaland Process Engineering, 2006.

12Hough, W. J., Autonomous Aerobatic Flight of a Fixed-Wing Unmanned Aerial Vehicle, Master’s thesis, Stellenbosch University, Departmentof Electrical and Electronic Engineering, 2007.

13Gaum, D. R., Aggressive Flight Control Techniques for a Fixed-Wing Unmanned Aerial Vehicle, Master’s thesis, Stellenbosch University,Department of Electrical and Electronic Engineering, 2009.

14Johnson, E. N., Turbe, M. A., Wu, A. D., Kannan, S. K., and Neidhoefer, J. C., “Flight Test Results of Autonomous Fixed-Wing UAVTransitions to and from Stationary Hover,” AIAA Paper 2006-6775, AIAA Guidance, Navigation, and Control Conference, Keystone, Colorado,Aug. 2006.

15Johnson, E. N., Wu, A. D., Neidhoefer, J. C., Kannan, S. K., and Turbe, M. A., “Test Results of Autonomous Airplane Transitions BetweenSteady-Level and Hovering Flight,” Journal of Guidance, Control, and Dynamics, Vol. 31, No. 2, 2008, pp. 358–370.

16Johnson, E. N. and Oh, S. M., “Relative Motion Estimation for Vision-based Formation Flight using Unscented Kalman Filter,” AIAA Paper2007-6866, AIAA Guidance, Navigation, and Control Conference, Hilton Head, South Carolina, Aug. 2007.

17Shah, S. I. A., Kanna, S. K., and Johnson, E. N., “Motion Estimation for Obstacle Detection and Avoidance Using a Single Camera forUAVs/Robots,” AIAA Paper 2010-7569, AIAA Guidance, Navigation, and Control Conference, Toronto, Ontario, Canada, Aug. 2010.

18Saha, B., Koshimoto, E., Quach, C. C., Hogge, E. F., Strom, T. H., Hill, B. L., Vazquez, S. L., and Goebel, K., “Battery Health ManagementSystem for Electric UAVs,” IEEE Aerospace Conference, Big Sky, Montana, 2011.

19Uhlig, D., Bhamidipati, K. K., and Neogi, N., “Safety and Reliability Within UAV Construction,” Proceedings of the IEEE Digital AvionicsSystem Conference, Portland, Oregon, 2006.

20Bhamidipati, K. K., Uhlig, D., and Neogi, N., “Engineering Safety and Reliability into UAV Systems: Mitigating the Ground Impact Hazard,”AIAA Paper 2007-6510, AIAA Guidance, Navigation, and Control Conference, Hilton Head, South Carolina, Aug. 2007.

21Ortiz, A. and Neogi, N., “A Dynamic Threshold Approach to Fault Detection in Unihabited Aerial Vehicles,” AIAA Paper 2008-7420, AIAAGuidance, Navigation, and Control Conference, Honolulu, Hawaii, Aug. 2008.

22Ortiz, A. and Neogi, N., “Color Optic Flow: A Computer Vision Approach for Object Detection on UAVs,” Proceedings of the IEEE DigitalAvionics System Conference, Portland, Oregon, 2006.

23Ortiz, A. and Neogi, N., “Object Detection and Avoidance Using Optical Techniques in Uninhabited Aerial Vehicles,” AIAA Paper 2007-6610, AIAA Guidance, Navigation, and Control Conference, Hilton Head, South Carolina, Aug. 2007.

24Danilov, C., Henderson, T. R., Goff, T., Kim, J. H., Macker, J., Weston, J., Neogi, N., Ortiz, A., and Uhlig, D., “Experiment and fielddemonstration of a 802.11-based ground-UAV mobile ad-hoc network,” Proceedings of the IEEE Military Communications Conference, Boston,Massachusetts, 2009.

25Leong, H. I., Keshmiri, S., and Jager, R., “Evaluation of a COTS Autopilot and Avionics System for UAVs,” AIAA Paper 2009-1963, AIAAInfotech@Aerospace, Seattle, Washington, April. 2009.

26Jager, R., Test and Evaluation of the Piccolo II Autopilot System on a One-Third Scale Yak-54, Master’s thesis, University of Kansas,Department of Aerospace Engineering, 2009.

27Esposito, J. F. and Keshmiri, S., “Rapid Hardware Interfacing and Software Development for Embedded Devices Using Simulink,” AIAAPaper 2010-3415, AIAA Infotech@Aerospace, Atlanta, Georgia, June 2010.

28Garcia, G. and Keshmiri, S., “Integrated Kalman Filter for a Flight Control System with Redundant Measurements,” AIAA Paper 2012-2499,AIAA Infotech@Aerospace, Garden Grove, California, June 2012.

29Keshmiri, S., Leong, E., Jager, R., and Hale, R., “Modeling and Simulation of the Yak-54 Scaled Unmanned Aerial Vehicle Using Parameterand System Identification,” AIAA Paper 2008-6900, AIAA Atmospheric Flight Mechanics Conference, Honolulu, Hawaii, Aug. 2008.

30Lykins, R. and Keshmiri, S., “System Identification of a COTS Autopilot System Using Flight Test Data,” AIAA Paper 2010-7648, AIAAAtmospheric Flight Mechanics Conference, Toronto, Ontario, Canada, Aug. 2010.

17 of 18


31Lykins, R. and Keshmiri, S., “Modal Analysis of 1/3-Scale Yak-54 Aircraft Through Simulation and Flight Testing,” AIAA Paper 2011-6443,AIAA Atmospheric Flight Mechanics Conference, Portland, Oregon, Aug. 2011.

32Jones, V. A. and Keshmiri, S., “First Flight Risk Mitigation Related to Pilot Unfamiliarity with a New UAV Platform,” AIAA Paper 2010-3489, AIAA Infotech@Aerospace, Atlanta, Georgia, June 2010.

33Sebes, J., VanSkike, W., Williams, M., McCandless, S. E., Stastny, T., Worden, G., and Brunkhorst, N., “Flight Testing and Evaluation ofthe Structural Response to Flight Loads of a Small Scale Unmanned Aerial System,” AIAA Paper 2012-2498, AIAA Infotech@Aerospace, GardenGrove, California, June 2012.

34Bowman, A., Barnes, G., and Keshmiri, S., “An Empirical Method for Estimating Moments of Inertia of Light Unmanned Air Vehicles,”AIAA Paper 2012-2598, AIAA Infotech@Aerospace, Garden Grove, California, June 2012.

35Selig, M. S., “Modeling Full-Envelope Aerodynamics of Small UAVs in Realtime,” AIAA Paper 2010-7635, AIAA Atmospheric FlightMechanics Conference, Toronto, Ontario, Canada, Aug. 2010.

36Selig, M. S., “Modeling Propeller Aerodynamics and Slipstream Effects on Small UAVs in Realtime,” AIAA Paper 2010-7635, AIAAAtmospheric Flight Mechanics Conference, Toronto, Ontario, Canada, Aug. 2010.

37Dantsker, O. D., Johnson, M. J., Akce, A. A., and Bretl, T. W., “Development of a Fixed Wing Multi-Role Unmanned Aircraft VehicleResearch Testbed,” AIAA Paper 2012-0846, AIAA Aerospace Sciences Meeting, Nashville, Tennessee, Jan. 2012.

38Akce, A. A., Johnson, M. J., and Bretl, T. W., “Remote Teleoperation of an Unmanned Aircraft with a Brain-Machine Interface: Theory andPreliminary Results,” Proceedings of the IEEE International Conference on Robotics and Automation, Ankorage, Alaska, 2010.

39Akce, A. A., Johnson, M. J., Dantsker, O. D., and Bretl, T. W., “A Brain-Machine Interface to Navigate a Mobile Robot in a PlanarWorkspace: Enabling Humans to Fly Simulated Aircraft with EEG,” IEEE Transactions on Neural Systems & Rehabilitation Engineering, Vol. 21,No. 2, pp. 306–318.

40Monticello Model Masters, “Eli Field,” http://www.elifield.com, Accessed Nov. 2012.41Desert Aircraft, “DA-120,” http://www.desertaircraft.com/engines detail.php?Page=DA-120, Accessed Nov. 2012.42Aero-Model Inc., “Hacker Brushless A150-10,” http://www.aero-model.com/Hacker-Brushless-A150-10.aspx, Accessed Nov. 2012.43Horizon Hobby, “35% Extra 260 ARF,” http://www.hangar-9.com/Products/Default.aspx?ProdID=HAN1000, Accessed Nov. 2012.44Graph Tech R/C, “Hangar 9 35% Extra 300 Main Gear,” http://www.graphtechrcshop.com/product.sc?productId=49categoryId=6, Accessed

Nov. 2012.45DuBro Products, Inc, “Home–Du-Bro Products, Inc.” http://hobby.dubro.com/, Accessed Nov. 2012.46Horizon Hobby, “JR Radios,” http://www.jrradios.com/, Accessed Nov. 2012.472icrc, “XPO Co-op Pilot System,” http://www.2icrc.com/, Accessed Nov. 2012.48Quest Engineering & Development, Inc., “Smart-Fly,” http://www.smart-fly.com/, Accessed Nov. 2012.49Advanced Energy Tech, “Thunder Power RC,” http://thunderpowerrc.com/, Accessed Nov. 2012.50Mejzlik Modellbau, “Carbon propellers and spinners–Mejzlik Modellbau,” http://www.mejzlik.eu/, Accessed Nov. 2012.51EMCOTEC GmbH., “RC Electronic by EMCOTEC,” http://shop.rc-electronic.com/, Accessed Nov. 2012.52Eagle Tree Systems, LLC, “Eagle Tree R/C Telemetry,” http://www.eagletreesystems.com/, Accessed Nov. 2012.53Gumstix, Inc., “Gumstix small open source hardware,” http://www.gumstix.com/, Accessed Nov. 2012.54The Paparazzi Project, LLC, “Paparazzi,” http://paparazzi.enac.fr/, Accessed Nov. 2012.55Xsens Technologies B.V., “XSens, MTi-g,” http://www.xsens.com/en/general/mti-g/, Accessed Nov. 2012.56All Sensors Corp., “Minature Amplified Pressure Sensors,” http://www.allsensors.com/products/20-cmH2O-D1-4V-MINI/, Accessed Nov.

2012.57Digi International Inc., “XTend RF Modems–1 Watt/900 MHz Stand-Alone Radio Modems,” http://www.digi.com/products/wireless-

modems-peripherals/wireless-range-extenders-peripherals/xtendoverview/, Accessed Nov. 2012.58Castle Creations, Inc., “Castle Creations–CC BEC and CC BEC Pro,” http://www.castlecreations.com/products/ccbec.html, Accessed Nov.

2012.59Ragheb, A. M., Dantsker, O. D., and Selig, M. S., “Stall/Spin Flight Testing with a Subscale Aerobatic Aircraft,” AIAA Paper 2013-2806,

AIAA Applied Aerodynamics Conference, San Diego, California, June 2013.60Dantsker, O. D., Johnson, M. J., Selig, M. S., and Bretl, T. W., “UIUC Aero Testbed,” http://www.youtube.com/watch?v=Rn5fZ7zyChc,

June 2013.

18 of 18


Documents

Development of the UIUC Aero Testbed: A Large …...Development of the UIUC Aero Testbed: A Large-Scale Unmanned Electric Aerobatic Aircraft for Aerodynamics Research Or D. Dantsker,